Jaguar vs Jaguar Options

[10p6]

Interesting stuff here. The only thing I ever knew that was different between the revisions was the lack of Analogue control between them. This is cool. I wonder if that means one system revision would make a better dev unit over the other?

Interesting thought. All I know is the data tends to suggest that if you want to play 3D then the K is better, and if you want to play platform games or 68K based ones then the difference is so marginal that I would still choose K. :-)

[T2KFreeker]

Interesting stuff here. The only thing I ever knew that was different between the revisions was the lack of Analogue control between them. This is cool. I wonder if that means one system revision would make a better dev unit over the other?

Interesting thought. All I know is the data tends to suggest that if you want to play 3D then the K is better, and if you want to play platform games or 68K based ones then the difference is so marginal that I would still choose K. :-)

Yeah, but wouldn't the 68K chips be the same anyway? I could be wrong, sorry. I tried to learn programming for the Jaguar and all it did was give me a headache, so I have to eject and forget. Only reason I can bring this up is that in the NEC world, there are two different board and chip designs for the US Turbografx 16. One designed by NEC, which are the earlier ones, and then the later/Last ones developed by Turbo Technologies, which was Hudson US and NEC US. It lead to the systems being lighter and even though it had no direct effect on Hu Catd games, when you used the CD Add On with TTi released Turbografx systems, load times and graphical clarity were improved. This was also the chipset used/scaled down for the Turbo Duo. However, due to the Bad Capacitors that were passed around in the early 90's that kills the US Duo's and Turbo Express units and Game Gears, The TTi Turbos have a pretty high failure rate until the caps are replaced. Not that that has any real point as to the performance of the chips in the end. Anyway, after the blabbing, is there any improvement or whatnot in CD behavior? Clearer graphics or whatnot by model?

[10p6]

Interesting stuff here. The only thing I ever knew that was different between the revisions was the lack of Analogue control between them. This is cool. I wonder if that means one system revision would make a better dev unit over the other?

Interesting thought. All I know is the data tends to suggest that if you want to play 3D then the K is better, and if you want to play platform games or 68K based ones then the difference is so marginal that I would still choose K. :-)

Yeah, but wouldn't the 68K chips be the same anyway? I could be wrong, sorry. I tried to learn programming for the Jaguar and all it did was give me a headache, so I have to eject and forget. Only reason I can bring this up is that in the NEC world, there are two different board and chip designs for the US Turbografx 16. One designed by NEC, which are the earlier ones, and then the later/Last ones developed by Turbo Technologies, which was Hudson US and NEC US. It lead to the systems being lighter and even though it had no direct effect on Hu Catd games, when you used the CD Add On with TTi released Turbografx systems, load times and graphical clarity were improved. This was also the chipset used/scaled down for the Turbo Duo. However, due to the Bad Capacitors that were passed around in the early 90's that kills the US Duo's and Turbo Express units and Game Gears, The TTi Turbos have a pretty high failure rate until the caps are replaced. Not that that has any real point as to the performance of the chips in the end. Anyway, after the blabbing, is there any improvement or whatnot in CD behavior? Clearer graphics or whatnot by model?

The only thing I thought about with CD was with IS2 for Cart and CD across both models. I am also curious though if the CD unit makes any difference with normal Cart games, so I guess sometime I will make a new quad screen Doom version to see.

[jvas]

First off, that is really cool, thank you for putting it together!!! I'm not noticing the differences yet, but I'll play closer attention!!!

If you look at the one on the left you will see that a lot of the time it is 1 or more frames ahead of the one on the right, even though both demo's are identical, and both demos started at exactly the same time.

If the difference is constantly 1 frame, then I wouldn't say one of them is faster than the other. How many units did you compare?

[10p6]

I check both of them and non are connected (No jumpers, resistors) What exactly does the R137 jumper do? Any documentation anywhere about all this stuff?

Bummer, that would have been an interesting find. I've only seen them non-connected.

I don't know if R137 is documented anywhere. I found out about it reading the Jaguar ROM code. When jumpered, R137 increases the size of a memory page to 4KB from 2KB. Larger pages slightly boost memory performance, by reducing page miss stalls.

You can't change the jumper without also changing the memory chips. It may be that no Jaguar ever used 4KB-page memory chips. (They are known as "512 column" or 512x512x16 in the datasheets - while most Jaguars seem to have 256 column or 1024x256x16 chips.)

This is the only jumper that changes memory performance. Otherwise, the Jaguar treats all memory chips the same. Even if they are faster, the Jaguar runs them at the usual speed.

Here are some other possible differences:

It might have something to do with the way the games are written. If the game doesn't sync to vblank when the demo starts, each run might be slightly different. It sounds like Sky Hammer might have this problem.

It could be down to different ROMs. Early Jaguar ROMs had slightly different video timings, didn't support PAL, and didn't spend extra time initializing memory to all zeros during startup. All these changes might affect demo timing or even behavior in subtle ways.

Even if they have the same ROM, the Jaguar clock crystals aren't that precise, so it's possible one Jaguar is running at 59.94 fps and another is running at 59.93 fps. After 15-20 seconds of recording, this could cause your capture card to duplicate a frame - putting one Jaguar behind the other.

To really get to the bottom of this, you might need more precise equipment. Digital video grabbers, cameras, scalers, even TV sets made in the last 10 years "retime" analog video - resulting in frames being inserted or dropped now and then. Old capture cards from the 90s generally didn't retime since it wasn't practical yet (but anything USB will - retiming is part of the UVC spec). Even an old VCR with frame-advance could record true frame timing, instead of retiming frames like modern stuff does.

- KS

I tried putting a switch on the R137 to see what happens. If switched before the Jag is turned on then it will not boot. If it is switched when the Jag is on then it simply does nothing.

[kskunk]

I tried putting a switch on the R137 to see what happens. If switched before the Jag is turned on then it will not boot. If it is switched when the Jag is on then it simply does nothing.

That's correct. Those resistors (135-137,140) are only tested during boot. And R137 requires a newer kind of RAM chip, which may never have made it into the Jaguar. Since it would boost performance (but only slightly), they might have purposefully avoided switching the RAM. Performance changes can cause bugs.

- Ks

[10p6]

Here is another test which takes the Jaguar and Doom configurations to the max. This test not only compares the Doom auto demo being played on a K and M model Jaguar, but it also compares then with Doom being run from a SkunkBoard, and also it being accessed through the CD unit.

Doom Comparison Video (Note that every video is synced to start at the exact same frame)


In the video you will see that the SkunkBoard gives quite a performance boost in transferring data for games like Doom. Also note that even though not a huge increase in speed, the CD unit did run the demo slightly faster with a Standart Cart, and overall the K Model runs Doom faster than the M model in all configurations.


After seeing the test videos, and talking with KSkunk, it would appear that a 3D Jaguar game could have a significant speed boost if it was ran directly from a flash cart, especially if no decompression was used. So anyone up to develop an inexpensive paged flash cart with 6MB pages. (The idea is that one level or segment runs from one page at a time)

[JagChris]

That was pretty hard to make sense out of lol. Maybe you should of just split it into 5 or 6 different views playing at once.

[10p6]

That was pretty hard to make sense out of lol. Maybe you should of just split it into 5 or 6 different views playing at once.

You don't like the, 'Side by side by side by side by side by side by side by side' video. LOL

[JagChris]

That was pretty hard to make sense out of lol. Maybe you should of just split it into 5 or 6 different views playing at once.

You don't like the, 'Side by side by side by side by side by side by side by side' video. LOL

It was like playing Vid Grid, except with Doom!